(Oops, just realized I only replied to a part of your message. I'm
blaming it on my lack of sleep.)
On 14/09/2013 06:53, chris travers wrote:
2) If you're running in strict mode and you want to
insert/update/delete more than one row, things get a bit uglier; a wCTE
would work for some cases. If strict mode doesn't affect EXECUTE (see
point 1 above), that could work too. Or maybe there could be a new
command which runs a query, discards the results and ignores the number
of rows processed.
Yeah, I am worried about this one. I am concerned that if you can't disable on
a statement by statement basis, then you have a problem where you end up
removing the mode from the function and then it becomes a lot harder for
everyone maintaining the function to have a clear picture of what is going on.
I am further worried that hacked ugly code ways around it will introduce
plenty
of other maintenance pain points that will be worse than what you are solving.
I completely agree. If you can't turn the option on globally, you're
bound to hit problems at some point.
My thinking currently is that this mode would require adding a new type
of statement (I'd call it PERFORM, but I know that one's been taken, so
please don't get hung up on that):
PERFORM UPDATE foo SET ..;
or
PERFORM SELECT foo(id) FROM ..;
This statement would allow you to run any statement in strict mode
without hitting the "more than one row processed" error message.
I'll be adding this to the open commitfest in hopes of getting some
feedback on this idea (I'm prepared to hear a lot of "you're crazy!"),
but feel free to comment away any time you please.
Well, I don't know if my feedback above is helpful, but there it is ;-)
Thanks for your input!
Regards,
Marko Tiikkaja
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
